jamescalam's picture
Create README.md
cca5e59
|
raw
history blame
335 Bytes
metadata
language:
  - en
pretty_name: Chunked Arxiv Papers for Llama 2

This dataset contains chunked extracts (of ~300 tokens) from papers related to (and including) the Llama 2 research paper. Related papers were identified by following a trail of references, extracting those papers, and repeating.